A proteomics sample pre-treatment device and a proteomics sample pre-treatment method

Through deep learning neural networks, the centrifugal process of proteomic samples is automatically controlled, which solves the problem of integration of sample processing and inconsistency of manual operation, and improves the efficiency and accuracy of proteomic samples processing.

CN118424809BActive Publication Date: 2024-11-22SHANGHAI ADICON CLINICAL LAB LNC
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410508137.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-04-25
Publication Date
2024-11-22
Estimated Expiration
2044-04-25

AI Technical Summary

Technical Problem

The lack of integrated methods during proteomic sample pretreatment results in sample loss and inconsistency in manual centrifugation operations, increasing experimental costs and time.

Method used

Using data processing and analysis algorithm based on deep learning neural network, the sample centrifugal state monitoring video is collected through the camera, local timing characteristics are extracted and automatic control instructions are generated to realize real-time monitoring and control of the sample centrifugal state.

Benefits of technology

It reduces the need for manual monitoring, improves the efficiency and accuracy of proteomic sample processing, and ensures consistency of centrifugal effect.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118424809B_ABST
    Figure CN118424809B_ABST
Patent Text Reader

Abstract

The present application discloses a proteomics sample pre-processing device and a processing method thereof, which obtains a sample centrifugal state monitoring video collected by a camera, and uses a data processing and analysis algorithm based on a deep learning neural network to process the sample centrifugal state monitoring video, so as to analyze and correlate the local time series characteristics of the centrifugal state, thereby automatically generating a control instruction on whether to stop the centrifugal operation based on the local time series correlation characteristics of the centrifugal state of each sample in the sample centrifugal state monitoring video. In this way, real-time monitoring and control of the sample centrifugal state can be achieved, reducing the need for manual monitoring, and improving the efficiency and accuracy of proteomics sample processing.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of intelligent processing, and more specifically, to a proteomics sample pre-processing device and a processing method thereof. Background Art

[0002] Proteomics samples refer to samples used for proteomics research, which can come from a variety of biological sources, including cells, tissues, body fluids, etc. Proteomics sample pretreatment is a very important step in proteomics research. Its purpose is to remove interfering substances, enrich target proteins, and convert samples into a form suitable for mass spectrometry analysis.

[0003] Currently, glycoproteomic analysis methods have not yet achieved a single integrated approach and still require several different pipette tip combinations for sample processing, or the proteins must be reduced and alkylated prior to subsequent enzymatic enrichment. These methods inevitably involve liquid transfer resulting in sample loss.

[0004] In response to the above technical problems, Chinese patent CN114509521A discloses a glycoproteomics sample pretreatment device and method, which can realize the integrated pre-enrichment, reductive alkylation, enzymatic hydrolysis, peptide desalting and complete glycopeptide enrichment of glycoprotein samples, solving the above-mentioned single problem. In the glycoproteomics sample pretreatment process, the glycoprotein sample to be tested needs to be dissolved in a lysate and acidified, and then added to an activated sample pretreatment device for centrifugation to remove solid particles and impurities in the sample and purify the glycoprotein sample to be tested. However, traditional centrifugation control is performed manually, which requires a lot of manpower and time, and there may be differences in the centrifugation operations of different operators, which not only increases the experimental cost but also leads to inconsistency in the centrifugation effect.

[0005] Therefore, an optimized proteomics sample preparation protocol is desired. Summary of the invention

[0006] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a proteomics sample pre-processing device and a processing method thereof, which obtains a sample centrifugal state monitoring video collected by a camera, and uses a data processing and analysis algorithm based on a deep learning neural network to process the sample centrifugal state monitoring video, so as to analyze and correlate the local temporal characteristics of the centrifugal state, thereby automatically generating a control instruction on whether to stop the centrifugal operation based on reasoning based on the local temporal correlation characteristics of the centrifugal state of each sample in the sample centrifugal state monitoring video. In this way, real-time monitoring and control of the sample centrifugal state can be achieved, reducing the need for manual monitoring, and improving the efficiency and accuracy of proteomics sample processing.

[0007] According to one aspect of the present application, a proteomics sample pre-treatment method is provided, comprising:

[0008] Using an organic solvent and a first acidic aqueous solution to sequentially activate, clean and balance the sample pretreatment device to obtain an activated sample pretreatment device;

[0009] Dissolving the glycoprotein sample to be tested in a lysis solution to obtain a lysis glycoprotein sample to be tested, and subjecting the lysis glycoprotein sample to acidification to obtain an acidified glycoprotein sample to be tested, wherein the lysis solution comprises a surfactant;

[0010] Adding the acidified glycoprotein sample to be tested into the activated sample pre-treatment device and subjecting it to centrifugal treatment, so that the glycoprotein sample to be tested is enriched on the mixed ion exchange resin to form a C18 extraction membrane;

[0011] Using a first organic solution to clean the surfactant on the C18 extraction membrane;

[0012] Adding a protein reducing agent and a protease solution to the activated sample pretreatment device in sequence to perform reduction, alkylation and enzymatic hydrolysis on the glycoprotein sample to be tested to obtain an enzymatic polypeptide;

[0013] Using a salt solution to transfer the enzymatically hydrolyzed polypeptide to the C18 extraction membrane, and using a second acidic aqueous solution to desalt the polypeptide to obtain a desalted enzymatically hydrolyzed polypeptide;

[0014] Using a second organic solution to transfer the desalted enzymatic polypeptide to a HILIC filler, and enriching the desalted enzymatic polypeptide for intact glycopeptides to obtain enriched intact glycopeptides;

[0015] Using a third acidic aqueous solution to elute the enriched intact glycopeptides to obtain intact glycopeptides;

[0016] The acidified glycoprotein sample to be tested is added to the activated sample pretreatment device and centrifuged to enrich the glycoprotein sample to be tested on the mixed ion exchange resin to form a C18 extraction membrane, comprising:

[0017] Obtaining a monitoring video of the centrifugal state of the sample collected by a camera;

[0018] After dividing the sample centrifugal state monitoring video into video segments to obtain a sequence of sample centrifugal state monitoring video segments, extracting sample centrifugal state time series correlation features from each sample centrifugal state monitoring video segment in the sequence of sample centrifugal state monitoring video segments to obtain a sequence of sample centrifugal state time series correlation feature graphs;

[0019] Passing each sample centrifugal state time series correlation feature graph in the sequence of the sample centrifugal state time series correlation feature graphs through a feature filter based on a gated switch attention network to obtain a sequence of purified sample centrifugal state time series correlation feature vectors;

[0020] The sequence of centrifugal state time series associated feature vectors of the purified sample is passed through a centrifugal state reasoner based on prior information to obtain a centrifugal state time series reasoning semantic feature vector as a centrifugal state time series reasoning semantic feature;

[0021] Based on the temporal reasoning semantic features of the centrifugal state, a control instruction is obtained.

[0022] According to another aspect of the present application, a proteomics sample pre-processing device is provided, wherein the proteomics sample pre-processing device is used to execute the aforementioned proteomics sample pre-processing method.

[0023] Compared with the prior art, the present application provides a proteomics sample pre-processing device and a processing method thereof, which obtains a sample centrifugal state monitoring video collected by a camera, and uses a data processing and analysis algorithm based on a deep learning neural network to process the sample centrifugal state monitoring video, so as to analyze and correlate the local time series characteristics of the centrifugal state, thereby automatically generating a control instruction on whether to stop the centrifugal operation based on the local time series correlation characteristics of the centrifugal state of each sample in the sample centrifugal state monitoring video. In this way, real-time monitoring and control of the sample centrifugal state can be achieved, reducing the need for manual monitoring, and improving the efficiency and accuracy of proteomics sample processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0025] Figure 1 The present invention is a flow chart of adding the acidified glycoprotein sample to be tested into the activated sample pretreatment device for centrifugal treatment in the proteomics sample pretreatment method according to an embodiment of the present application, so that the glycoprotein sample to be tested is enriched on a mixed ion exchange resin.

[0026] Figure 2The present invention is a schematic diagram of the structure of adding the acidified glycoprotein sample to be tested into the activated sample pretreatment device for centrifugation in the proteomics sample pretreatment method according to an embodiment of the present application, so that the glycoprotein sample to be tested is enriched on the mixed ion exchange resin.

[0027] Figure 3 The present invention is a flowchart of the method for preprocessing proteomics samples according to an embodiment of the present application, which comprises dividing the sample centrifugation state monitoring video into video segments to obtain a sequence of sample centrifugation state monitoring video segments, and then extracting sample centrifugation state time series correlation features from each sample centrifugation state monitoring video segment in the sequence of sample centrifugation state monitoring video segments to obtain a sequence of sample centrifugation state time series correlation feature graphs.

[0028] Figure 4 The present invention is a flowchart for training the sample centrifugal state temporal correlation feature extractor based on the three-dimensional convolutional neural network model, the feature filter based on the gated switch attention network, the centrifugal state reasoner based on prior information, and the centrifugal controller based on the classifier in the proteomics sample pretreatment method according to an embodiment of the present application. DETAILED DESCRIPTION

[0029] The embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments described herein, but rather these embodiments are provided for a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0030] It should be understood that the various steps described in the method embodiments of the present disclosure may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this respect.

[0031] In the description of the embodiments of the present disclosure, the term "including" and similar terms should be understood as open inclusion, that is, "including but not limited to". The term "based on" should be understood as "based at least in part on". The term "one embodiment" or "the embodiment" should be understood as "at least one embodiment". The terms "first", "second", etc. may refer to different or the same objects. Other explicit and implicit definitions may also be included below.

[0032] It should be noted that the modifications of "one" and "plurality" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0033] Proteomics samples refer to samples used for proteomics research, which can come from a variety of biological sources, including cells, tissues, body fluids, etc. Proteomics sample pretreatment is a very important step in proteomics research. Its purpose is to remove interfering substances, enrich target proteins, and convert samples into a form suitable for mass spectrometry analysis.

[0034] Currently, glycoproteomic analysis methods have not yet achieved a single integrated approach and still require several different pipette tip combinations for sample processing, or the proteins must be reduced and alkylated prior to subsequent enzymatic enrichment. These methods inevitably involve liquid transfer resulting in sample loss.

[0035] In view of the above technical problems, Chinese patent CN114509521A discloses a glycoproteomics sample pretreatment device and method, which comprises: using an organic solvent and a first acidic aqueous solution to sequentially activate, clean and balance a sample pretreatment device to obtain an activated sample pretreatment device; dissolving a glycoprotein sample to be tested in a lysis solution to obtain a lysis glycoprotein sample to be tested, and acidifying the lysis glycoprotein sample to be tested to obtain an acidified glycoprotein sample to be tested, wherein the lysis solution comprises a surfactant; adding the acidified glycoprotein sample to be tested to the activated sample pretreatment device for centrifugation, so that the glycoprotein sample to be tested is enriched on a mixed ion exchange resin; using a first organic solution to clean the glycoprotein sample bound to the mixed ion exchange resin; The surfactant on the C18 ion exchange resin and the C18 extraction membrane; the protein reducing agent and the protease solution are sequentially added to the activated sample pretreatment device to complete the reduction, alkylation and enzymatic hydrolysis of the glycoprotein sample to be tested to obtain enzymatic peptides; the enzymatic peptides are transferred to the C18 extraction membrane using a salt solution, and the peptides are desalted using a second acidic aqueous solution to obtain desalted enzymatic peptides; the desalted enzymatic peptides on the C18 extraction membrane are transferred to the HILIC filler using a second organic solution, and the desalted enzymatic peptides are enriched for complete glycopeptides to obtain enriched complete glycopeptides; and the enriched complete glycopeptides on the HILIC filler are eluted using a third acidic aqueous solution to obtain complete glycopeptides. The above method can realize the integrated whole process of pre-enrichment, reduction alkylation, enzymatic hydrolysis, peptide desalting and complete glycopeptide enrichment of glycoprotein samples, solving the above-mentioned single problem.

[0036] Among them, the acidified glycoprotein sample to be tested is added to the activated sample pretreatment device for centrifugal treatment to remove solid particles and impurities in the sample. In the step of purifying the glycoprotein sample to be tested, since centrifugation is traditionally controlled manually, it requires a lot of manpower and time, and the centrifugation operations of different operators may be different, which increases the experimental cost and also affects the inconsistency of the centrifugation effect.

[0037] Therefore, in view of the above technical problems, the technical concept of the present application is to obtain the monitoring video of the centrifugal state of the sample collected by the camera, and use the data processing and analysis algorithm based on the deep learning neural network to process the monitoring video of the centrifugal state of the sample, so as to analyze and correlate the local time series characteristics of the centrifugal state, so as to automatically generate a control instruction on whether to stop the centrifugal operation based on the local time series correlation characteristics of the centrifugal state of each sample in the monitoring video of the centrifugal state of the sample. In this way, real-time monitoring and control of the centrifugal state of the sample can be achieved, reducing the need for manual monitoring and improving the efficiency and accuracy of proteomics sample processing.

[0038] Figure 1 The present invention is a flow chart of adding the acidified glycoprotein sample to be tested into the activated sample pretreatment device for centrifugal treatment in the proteomics sample pretreatment method according to an embodiment of the present application, so that the glycoprotein sample to be tested is enriched on a mixed ion exchange resin. Figure 2 The schematic diagram of the structure of the proteomics sample pretreatment method according to the embodiment of the present application is to add the acidified glycoprotein sample to be tested into the activated sample pretreatment device for centrifugation, so that the glycoprotein sample to be tested is enriched on the mixed ion exchange resin. Figure 1 He Ru Figure 2As shown, the acidified glycoprotein sample to be tested is added to the activated sample pretreatment device for centrifugal treatment, so that the glycoprotein sample to be tested is enriched on the mixed ion exchange resin, including: S110, obtaining a sample centrifugal state monitoring video collected by a camera; S120, dividing the sample centrifugal state monitoring video into video segments to obtain a sequence of sample centrifugal state monitoring video segments, and then extracting sample centrifugal state time series correlation features from each sample centrifugal state monitoring video segment in the sequence of sample centrifugal state monitoring video segments to obtain a sequence of sample centrifugal state time series correlation feature graphs; S130, passing each sample centrifugal state time series correlation feature graph in the sequence of sample centrifugal state time series correlation feature graphs through a feature filter based on a gated switch attention network to obtain a sequence of purified sample centrifugal state time series correlation feature vectors; S140, passing the sequence of purified sample centrifugal state time series correlation feature vectors through a centrifugal state reasoner based on prior information to obtain a centrifugal state time series reasoning semantic feature vector as a centrifugal state time series reasoning semantic feature; and, S150, obtaining a control instruction based on the centrifugal state time series reasoning semantic feature.

[0039] In step S110, a monitoring video of the centrifugal state of the sample captured by the camera is obtained. It should be understood that considering that sample centrifugation is one of the commonly used operating steps in biological experiments. The monitoring video of the centrifugal state of the sample reflects various characteristic information during the sample centrifugation process, such as the position of the sample, the centrifugation time, etc. Based on this, in the technical solution of the present application, the monitoring video of the centrifugal state of the sample captured by the camera is obtained and processed and analyzed, and the state changes of the sample during the centrifugation process can be intuitively observed, so as to detect abnormal situations in time, thereby realizing automatic monitoring and control of the centrifugation process, helping experimenters to monitor the experimental process, and ensuring the accuracy and reliability of experimental data.

[0040] In step S120, the sample centrifugal state monitoring video is segmented into video segments to obtain a sequence of sample centrifugal state monitoring video segments, and then the sample centrifugal state time series correlation feature extraction is performed on each sample centrifugal state monitoring video segment in the sequence of sample centrifugal state monitoring video segments to obtain a sequence of sample centrifugal state time series correlation feature graphs. Figure 3 The present invention is a flow chart of the method for pre-processing a proteomics sample according to an embodiment of the present application, wherein the sample centrifugal state monitoring video is segmented into video segments to obtain a sequence of sample centrifugal state monitoring video segments, and then the sample centrifugal state time series correlation feature extraction is performed on each sample centrifugal state monitoring video segment in the sequence of sample centrifugal state monitoring video segments to obtain a sequence of sample centrifugal state time series correlation feature graphs. Specifically, in the embodiment of the present application, Figure 3As shown, after the sample centrifugal state monitoring video is divided into video segments to obtain a sequence of sample centrifugal state monitoring video segments, the sample centrifugal state time series correlation features of each sample centrifugal state monitoring video segment in the sequence of sample centrifugal state monitoring video segments are extracted to obtain a sequence of sample centrifugal state time series correlation feature graphs, including: S210, dividing the sample centrifugal state monitoring video into video segments to obtain a sequence of sample centrifugal state monitoring video segments; and, S220, respectively passing each sample centrifugal state monitoring video segment in the sequence of sample centrifugal state monitoring video segments through a sample centrifugal state time series correlation feature extractor based on a three-dimensional convolutional neural network model to obtain a sequence of sample centrifugal state time series correlation feature graphs.

[0041] Specifically, in step S210, the sample centrifugal state monitoring video is segmented into video segments to obtain a sequence of the sample centrifugal state monitoring video segments. Accordingly, considering that the long-term sample centrifugal state monitoring video may contain a large amount of information, processing and analyzing the entire sample centrifugal state monitoring video will increase the computational complexity and time cost. Therefore, in order to analyze and understand the local time segments of the sample centrifugal state monitoring video more finely, in the technical solution of the present application, the sample centrifugal state monitoring video is segmented into video segments to obtain a sequence of the sample centrifugal state monitoring video segments. That is, by segmenting the sample centrifugal state monitoring video into small segments, each small segment represents a sample centrifugal state monitoring video of a local time segment, so that the sample centrifugal state monitoring video of each local time segment can be more conveniently processed and analyzed, such as the position of the sample at different time points, the rotation speed and other information or the abnormal conditions during the sample centrifugal process, such as sample shedding, uneven centrifugation and other problems, because the abnormality often occurs within a certain time period, so as to achieve a more refined monitoring and analysis of the sample centrifugal state, and improve the accuracy of the judgment of the subsequent centrifugal operation instructions.

[0042] Specifically, in step S220, each sample centrifugal state monitoring video clip in the sequence of the sample centrifugal state monitoring video clips is respectively subjected to a sample centrifugal state temporal correlation feature extractor based on a three-dimensional convolutional neural network model to obtain a sequence of the sample centrifugal state temporal correlation feature graphs. It should be understood that, considering that each sample centrifugal state monitoring video clip in the sequence of the sample centrifugal state monitoring video clips will continuously change the centrifugal state over time, such as the position and rotation state of the sample, that is, the state of the sample during the centrifugal process is temporally correlated. Therefore, in order to respectively capture the temporal state change characteristics of each sample centrifugal state monitoring video clip during the centrifugal process, so as to better understand the temporal feature change rules of each sample centrifugal fragment process, in the technical solution of the present application, each sample centrifugal state monitoring video clip in the sequence of the sample centrifugal state monitoring video clip is respectively subjected to a sample centrifugal state temporal correlation feature extractor based on a three-dimensional convolutional neural network model to respectively capture and mine the change feature information of each sample centrifugal state monitoring video clip on the sample centrifugal state in the local time segment, so as to obtain a sequence of sample centrifugal state temporal correlation feature graphs.

[0043] In step S130, each sample centrifugal state time series correlation feature graph in the sequence of the sample centrifugal state time series correlation feature graph is passed through a feature filter based on a gated switch attention network to obtain a sequence of purified sample centrifugal state time series correlation feature vectors. Accordingly, considering that the sample centrifugal state foreground of each sample centrifugal state time series correlation feature graph in the sequence of the sample centrifugal state time series correlation feature graph has many common features, and there is also some irrelevant background information. Therefore, in order to focus on and filter out the most important foreground feature information about the local time series correlation feature of the sample centrifugal state in each sample centrifugal state time series correlation feature graph, so as to reduce the interference and influence of background information, in the technical solution of the present application, each sample centrifugal state time series correlation feature graph in the sequence of the sample centrifugal state time series correlation feature graph is passed through a feature filter based on a gated switch attention network to obtain a sequence of purified sample centrifugal state time series correlation feature vectors. It is worth mentioning that the gated switch attention network can dynamically adjust the focus of the key area of ​​the input data by learning the importance of the centrifugal state features in different video clips, so as to more accurately identify and analyze the centrifugal state content and time series changes. It should be understood that by processing each sample centrifugation state time-series correlation feature graph in the sequence of the sample centrifugation state time-series correlation feature graphs through the feature filter based on the gated switch attention network, the model can focus on the key foreground feature information about the sample centrifugation state in each sample centrifugation state time-series correlation feature graph, so as to better screen out the important feature information and time-series change information related to the centrifugation state of each sample centrifugation state time-series correlation feature graph, thereby obtaining a more representative and effective purified sample centrifugation state time-series correlation feature vector sequence.

[0044] Specifically, in an embodiment of the present application, each sample centrifugal state timing correlation feature graph in the sequence of the sample centrifugal state timing correlation feature graph is passed through a feature filter based on a gated switch attention network to obtain a sequence of purified sample centrifugal state timing correlation feature vectors, including: using the feature filter based on the gated switch attention network to process each sample centrifugal state timing correlation feature graph in the sequence of the sample centrifugal state timing correlation feature graph with the following feature screening formula to obtain the sequence of purified sample centrifugal state timing correlation feature vectors; wherein the feature screening formula is:

[0045]

[0046]

[0047] Among them, v i (x) and v j (x)are the i-th and j-th sample centrifugal state time series association feature vectors in the sequence of sample centrifugal state time series association feature vectors after each sample centrifugal state time series association feature graph is expanded, L is the scale of each sample centrifugal state time series association feature vector in the sequence of sample centrifugal state time series association feature vectors, V k represents the sequence of the time-series associated feature vectors of the centrifugal state of the sample, M represents the number of feature vectors in the sequence of the time-series associated feature vectors of the centrifugal state of the sample -1, D i Represents the feature value of each position in the local temporal correlation semantic difference feature vector of the centrifugal state of the sample, N s represents the number of eigenvalues ​​in the local temporal association semantic difference feature vector of the centrifugal state of the sample, V represents each purified sample centrifugal state temporal association feature vector in the set of purified sample centrifugal state temporal association feature vectors, exp(·) represents an exponential function with the natural constant e as the base, and log represents a logarithmic function value with 2 as the base.

[0048] In step S140, the sequence of the centrifugal state time series associated feature vectors of the purified samples is passed through a centrifugal state reasoner based on prior information to obtain a centrifugal state time series reasoning semantic feature vector as a centrifugal state time series reasoning semantic feature. It should be understood that, considering that there are certain differences in the centrifugal state time series associated feature vectors of the purified samples in the sequence of the centrifugal state time series associated feature vectors, and there is mutual influence and association between the centrifugal state features of different samples. Therefore, in order to perform global reasoning on the sequence of the centrifugal state time series associated feature vectors of the purified samples to obtain an optimal centrifugal state reasoning feature, and to provide support and assistance for subsequent centrifugal state control, in the technical solution of the present application, the sequence of the centrifugal state time series associated feature vectors of the purified samples is passed through a centrifugal state reasoner based on prior information to obtain a centrifugal state time series reasoning semantic feature vector. It should be understood that the centrifugal state reasoner based on prior information generates a larger disturbance near a specific feature vector by introducing a Gaussian distribution as prior information, which can help the algorithm to jump out of the local extreme value more easily and avoid falling into a local optimal solution and missing the opportunity of the global optimal solution. That is to say, by performing global reasoning on the differences between the time-series correlation feature vectors of the centrifugal states of the purified samples in the sequence of the time-series correlation feature vectors of the centrifugal states of the purified samples based on prior information, that is, updating the centrifugal state feature of the subsequent sample based on the centrifugal state feature of the previous sample of the centrifugal state temporal correlation feature of the centrifugal state of the purified sample, this can more accurately grasp the temporal change law and feature evolution of the sample during the centrifugation process, thereby more accurately capturing the time-series correlation feature of the centrifugal state of the purified samples about the time-series correlation feature of the centrifugal state of the sample in the last time period, so as to obtain the centrifugal state temporal reasoning semantic feature vector containing prediction information.

[0049] Specifically, in an embodiment of the present application, the sequence of the centrifugal state time series associated feature vectors of the purified sample is passed through a centrifugal state reasoner based on prior information to obtain a centrifugal state time series reasoning semantic feature vector as a centrifugal state time series reasoning semantic feature, including: using the centrifugal state reasoner based on prior information to process the sequence of the centrifugal state time series associated feature vectors of the purified sample using the following reasoning formula to obtain the centrifugal state time series reasoning semantic feature vector; wherein the reasoning formula is:

[0050]

[0051] Wherein, p1 and p2 are weight hyperparameters, f1 and f2 are hyperparameters that use a Gaussian distribution random number function with a mean of 0 and a variance of 1 as Gaussian distribution function coefficients, V1 is the last position of the purified sample centrifugal state time series correlation feature vector in the sequence of the purified sample centrifugal state time series correlation feature vector, V rand is the purified sample centrifugal state time series correlation feature vector at a random position in the sequence of the purified sample centrifugal state time series correlation feature vectors, V * is the mean feature vector of the sequence of time-series associated feature vectors of the centrifugal state of the purified sample, and are vector subtraction and vector addition respectively, and V is the centrifugal state temporal reasoning semantic feature vector.

[0052] In step S150, a control instruction is obtained based on the centrifugal state time series reasoning semantic features. Specifically, in the embodiment of the present application, the control instruction is obtained based on the centrifugal state time series reasoning semantic features, including: passing the centrifugal state time series reasoning semantic feature vector through a centrifugal controller based on a classifier to obtain a control instruction, and the control instruction is used to indicate whether to stop the centrifugal operation. That is, the centrifugal state time series reasoning semantic features obtained by reasoning based on prior information are used to classify and process the sequence of the purified sample centrifugal state time series association feature vectors, so as to automatically generate a control instruction on whether to stop the centrifugal operation based on the local time series association features of the centrifugal state of each sample in the sample centrifugal state monitoring video. In this way, real-time monitoring and control of the sample centrifugal state can be achieved, the need for manual monitoring can be reduced, and the efficiency and accuracy of proteomics sample processing can be improved.

[0053] It is worth mentioning that those skilled in the art should be aware that before applying a deep neural network model for inference, the deep neural network model must be trained so that the deep neural network can implement specific functional capabilities.

[0054] Specifically, in an embodiment of the present application, a training step is also included: for training the sample centrifugal state temporal correlation feature extractor based on the three-dimensional convolutional neural network model, the feature filter based on the gated switch attention network, the centrifugal state reasoner based on prior information, and the centrifugal controller based on the classifier.

[0055] Figure 4 The present invention is a flowchart of training the sample centrifugal state temporal correlation feature extractor based on the three-dimensional convolutional neural network model, the feature filter based on the gated switch attention network, the centrifugal state reasoner based on prior information, and the centrifugal controller based on the classifier in the proteomics sample pre-processing method according to the embodiment of the present application. Figure 4 As shown, the training step includes: S310, obtaining training data, wherein the training data includes obtaining a centrifugal state monitoring video of the training sample collected by a camera, and the true value of whether the centrifugal operation is stopped; S320, dividing the centrifugal state monitoring video of the training sample into video segments to obtain a sequence of centrifugal state monitoring video segments of the training sample; S330, passing each centrifugal state monitoring video segment of the sequence of centrifugal state monitoring video segments of the training sample through the sample centrifugal state temporal association feature extractor based on the three-dimensional convolutional neural network model to obtain a sequence of centrifugal state temporal association feature graphs of the training sample; S340, passing each centrifugal state temporal association feature graph of the training sample in the sequence of centrifugal state temporal association feature graphs through the feature filter based on the gated switch attention network to obtain a centrifugal state of the purified training sample. S350, passing the sequence of training purified sample centrifugal state time-series correlation feature vectors through the centrifugal state reasoner based on prior information to obtain training centrifugal state time-series reasoning semantic feature vectors; S360, correcting the training centrifugal state time-series reasoning semantic feature vectors to obtain corrected training centrifugal state time-series reasoning semantic feature vectors; S370, passing the corrected training centrifugal state time-series reasoning semantic feature vectors through the classifier-based centrifugal controller to obtain a classification loss function value; and, S380, based on the classification loss function and through back propagation of gradient descent, training the sample centrifugal state time-series correlation feature extractor based on the three-dimensional convolutional neural network model, the feature filter based on the gated switch attention network, the centrifugal state reasoner based on prior information and the centrifugal controller based on the classifier.

[0056] Specifically, in step S360, the training centrifugal state temporal reasoning semantic feature vector is corrected to obtain the corrected training centrifugal state temporal reasoning semantic feature vector. It should be understood that in the technical solution of the present application, each training purified sample centrifugal state temporal association feature vector in the sequence of the training purified sample centrifugal state temporal association feature vector respectively represents the purified feature representation of the sample centrifugal state video semantic temporal association feature after feature screening by the feature filter based on the gated switch attention network. When the sequence of the training purified sample centrifugal state temporal association feature vector passes through the centrifugal state reasoner based on prior information, considering that the centrifugal state reasoner based on prior information introduces Gaussian distribution as prior information to perform centrifugal state feature reasoning, and at the same time, a training purified sample centrifugal state temporal association feature vector randomly selected from the sequence of the training purified sample centrifugal state temporal association feature vector is used as a reference anchor point, which will make the generalization ability of the training centrifugal state temporal reasoning semantic feature vector stronger, but it will also make the training centrifugal state temporal reasoning semantic feature vector as a feature distribution as a whole have local feature distribution discreteness due to the introduction of random reference anchor points. That is, the training centrifugal state temporal reasoning semantic feature vector has rich geometric shape details in the high-dimensional image semantic feature space, so that there is a distortion in the category representation when the training centrifugal state temporal reasoning semantic feature vector is classified by the classifier to map to the class probability domain, that is, the category representation of the training centrifugal state temporal reasoning semantic feature vector for the classifier's predetermined category is affected, thereby reducing the accuracy of the classification result. Based on this, the present application corrects the training centrifugal state temporal reasoning semantic feature vector each time the training centrifugal state temporal reasoning semantic feature vector is classified iteratively by the classifier.

[0057] Specifically, in an embodiment of the present application, the training centrifugal state temporal reasoning semantic feature vector is corrected to obtain a corrected training centrifugal state temporal reasoning semantic feature vector, including: correcting the training centrifugal state temporal reasoning semantic feature vector based on the following correction formula to obtain a corrected training centrifugal state temporal reasoning semantic feature vector; wherein the correction formula is:

[0058]

[0059] Wherein, V is the training centrifugal state temporal reasoning semantic feature vector, p is the class probability value obtained by the classifier for the training centrifugal state temporal reasoning semantic feature vector V, and v i is the i-th eigenvalue of the training centrifugal state temporal reasoning semantic feature vector V, is the mean of all eigenvalues ​​of the training centrifugal state temporal reasoning semantic feature vector V, ‖V‖1 represents the norm of the training centrifugal state temporal reasoning semantic feature vector V, and α is a weight hyperparameter, vi ′ is the i-th eigenvalue of the corrected training eccentric state temporal reasoning semantic feature vector.

[0060] Specifically, while maintaining the geometric shape details of the high-dimensional feature manifold under the class probability mapping of the training centrifugal state temporal reasoning semantic feature vector V, the shape attribute editing is performed on the feature value details relative to the variational representation of the feature set of the training centrifugal state temporal reasoning semantic feature vector V through the distortion consultation of the class probability, thereby mapping the class probability distortion of the training centrifugal state temporal reasoning semantic feature vector V to the latent class space feature representation, and then consulting fusion is performed by supplementing the basic low-rank constrained latent inversion representation to bridge the gap between the edited consulting representation and the original geometric shape details, thereby improving the category representation of the high-dimensional features of the training centrifugal state temporal reasoning semantic feature vector V, thereby improving the accuracy of the classification results. In this way, real-time monitoring and control of the centrifugal state of the sample can be achieved, reducing the need for manual monitoring, and improving the efficiency and accuracy of proteomics sample processing.

[0061] In summary, the proteomics sample pretreatment method based on the embodiment of the present application is explained, which obtains the sample centrifugal state monitoring video collected by the camera, and uses the data processing and analysis algorithm based on the deep learning neural network to process the sample centrifugal state monitoring video, so as to analyze and correlate the local time series characteristics of the centrifugal state, thereby automatically generating a control instruction on whether to stop the centrifugal operation based on the local time series correlation characteristics of the centrifugal state of each sample in the sample centrifugal state monitoring video. In this way, real-time monitoring and control of the sample centrifugal state can be achieved, reducing the need for manual monitoring, and improving the efficiency and accuracy of proteomics sample processing.

[0062] Specifically, a proteomics sample pre-processing device is also provided, and the proteomics sample pre-processing device is used to execute the proteomics sample pre-processing method mentioned above.

[0063] The foregoing is merely an example of the principles of the present disclosure, and various modifications may be made by those skilled in the art without departing from the scope of the present disclosure. The above embodiments are presented for the purpose of illustration and not limitation. The present disclosure may also take many forms other than those explicitly described herein. Therefore, it is emphasized that the present disclosure is not limited to the explicitly disclosed methods and apparatus, but is intended to include variations and modifications within the spirit and scope of the appended claims.

Claims

1. A proteomics sample pretreatment method, comprising: Using an organic solvent and a first acidic aqueous solution to sequentially activate, clean and balance the sample pretreatment device to obtain an activated sample pretreatment device; Dissolving the glycoprotein sample to be tested in a lysis solution to obtain a lysis glycoprotein sample to be tested, and subjecting the lysis glycoprotein sample to acidification to obtain an acidified glycoprotein sample to be tested, wherein the lysis solution comprises a surfactant; Adding the acidified glycoprotein sample to be tested into the activated sample pre-treatment device and subjecting it to centrifugal treatment, so that the glycoprotein sample to be tested is enriched on the mixed ion exchange resin to form a C18 extraction membrane; Using a first organic solution to clean the surfactant on the C18 extraction membrane; Adding a protein reducing agent and a protease solution to the activated sample pretreatment device in sequence to perform reduction, alkylation and enzymatic hydrolysis on the glycoprotein sample to be tested to obtain an enzymatic polypeptide; Using a salt solution to transfer the enzymatically hydrolyzed polypeptide to the C18 extraction membrane, and using a second acidic aqueous solution to desalt the polypeptide to obtain a desalted enzymatically hydrolyzed polypeptide; Using a second organic solution to transfer the desalted enzymatic polypeptide to a HILIC filler, and enriching the desalted enzymatic polypeptide for intact glycopeptides to obtain enriched intact glycopeptides; Using a third acidic aqueous solution to elute the enriched intact glycopeptides to obtain intact glycopeptides; The acidified glycoprotein sample to be tested is added to the activated sample pretreatment device and centrifuged to enrich the glycoprotein sample to be tested on the mixed ion exchange resin to form a C18 extraction membrane, characterized in that it includes: Obtaining a monitoring video of the centrifugal state of the sample collected by a camera; After dividing the sample centrifugal state monitoring video into video segments to obtain a sequence of sample centrifugal state monitoring video segments, extracting sample centrifugal state time series correlation features from each sample centrifugal state monitoring video segment in the sequence of sample centrifugal state monitoring video segments to obtain a sequence of sample centrifugal state time series correlation feature graphs; Passing each sample centrifugal state time series correlation feature graph in the sequence of the sample centrifugal state time series correlation feature graphs through a feature filter based on a gated switch attention network to obtain a sequence of purified sample centrifugal state time series correlation feature vectors; The sequence of centrifugal state time series associated feature vectors of the purified sample is passed through a centrifugal state reasoner based on prior information to obtain a centrifugal state time series reasoning semantic feature vector as a centrifugal state time series reasoning semantic feature; Based on the temporal reasoning semantic features of the centrifugal state, a control instruction is obtained; The method of passing the sequence of the centrifugal state time series associated feature vectors of the purified samples through a centrifugal state reasoner based on prior information to obtain a centrifugal state time series reasoning semantic feature vector as a centrifugal state time series reasoning semantic feature comprises: using the centrifugal state reasoner based on prior information to process the sequence of the centrifugal state time series associated feature vectors of the purified samples with the following reasoning formula to obtain the centrifugal state time series reasoning semantic feature vector; Among them, the inference formula is: Wherein, p1 and p2 are weight hyperparameters, f1 and f2 are hyperparameters that use a Gaussian distribution random number function with a mean of 0 and a variance of 1 as Gaussian distribution function coefficients, V1 is the last position of the purified sample centrifugal state time series correlation feature vector in the sequence of the purified sample centrifugal state time series correlation feature vector, V rand is a purified sample centrifugal state time series associated feature vector at a random position in the sequence of purified sample centrifugal state time series associated feature vectors, V* is a mean feature vector of the sequence of purified sample centrifugal state time series associated feature vectors, and are vector subtraction and vector addition respectively, V is the centrifugal state temporal reasoning semantic feature vector; The method of obtaining a sequence of purified sample centrifugal state time series correlation feature vectors by subjecting each sample centrifugal state time series correlation feature graph in the sequence of sample centrifugal state time series correlation feature graphs to a feature filter based on a gated switch attention network, comprises: using the feature filter based on a gated switch attention network to process each sample centrifugal state time series correlation feature graph in the sequence of sample centrifugal state time series correlation feature graphs using the following feature screening formula to obtain a sequence of purified sample centrifugal state time series correlation feature vectors; Among them, the feature screening formula is: Among them, v i (x) and v j (x) are the i-th and j-th sample centrifugal state time series association feature vectors in the sequence of sample centrifugal state time series association feature vectors after each sample centrifugal state time series association feature graph is expanded, L is the scale of each sample centrifugal state time series association feature vector in the sequence of sample centrifugal state time series association feature vectors, V k represents the sequence of the time-series associated feature vectors of the centrifugal state of the sample, M represents the number of feature vectors in the sequence of the time-series associated feature vectors of the centrifugal state of the sample -1, D i Represents the feature value of each position in the local temporal correlation semantic difference feature vector of the centrifugal state of the sample, N s represents the number of eigenvalues ​​in the local temporal association semantic difference feature vector of the centrifugal state of the sample, V represents each purified sample centrifugal state temporal association feature vector in the set of purified sample centrifugal state temporal association feature vectors, exp(·) represents an exponential function with the natural constant e as the base, and log represents a logarithmic function value with 2 as the base.

2. The proteomics sample pretreatment method according to claim 1, characterized in that: The sample centrifugal state monitoring video is segmented into video segments to obtain a sequence of sample centrifugal state monitoring video segments, and then the sample centrifugal state time series correlation feature extraction is performed on each sample centrifugal state monitoring video segment in the sequence of sample centrifugal state monitoring video segments to obtain a sequence of sample centrifugal state time series correlation feature graphs, including: Segmenting the sample centrifugal state monitoring video into video segments to obtain a sequence of the sample centrifugal state monitoring video segments; Each sample centrifugal state monitoring video clip in the sequence of the sample centrifugal state monitoring video clips is respectively passed through a sample centrifugal state time series correlation feature extractor based on a three-dimensional convolutional neural network model to obtain a sequence of the sample centrifugal state time series correlation feature graphs.

3. The proteomics sample pretreatment method according to claim 2, characterized in that: Based on the centrifugal state temporal reasoning semantic feature, a control instruction is obtained, including: passing the centrifugal state temporal reasoning semantic feature vector through a classifier-based centrifugal controller to obtain a control instruction, wherein the control instruction is used to indicate whether to stop the centrifugal operation.

4. The proteomics sample pretreatment method according to claim 3, characterized in that: It also includes a training step: for training the sample centrifugal state temporal correlation feature extractor based on the three-dimensional convolutional neural network model, the feature filter based on the gated switch attention network, the centrifugal state reasoner based on prior information and the centrifugal controller based on the classifier.

5. The proteomics sample pretreatment method according to claim 4, characterized in that: The training step comprises: Acquiring training data, wherein the training data includes acquiring a centrifugal state monitoring video of a training sample captured by a camera, and a true value of whether to stop the centrifugal operation; Segmenting the training sample centrifugal state monitoring video into video segments to obtain a sequence of training sample centrifugal state monitoring video segments; Each of the training sample centrifugal state monitoring video clips in the sequence of the training sample centrifugal state monitoring video clips is respectively passed through the sample centrifugal state time series correlation feature extractor based on the three-dimensional convolutional neural network model to obtain a sequence of training sample centrifugal state time series correlation feature graphs; Passing each of the training sample centrifugal state time series correlation feature graphs in the sequence of the training sample centrifugal state time series correlation feature graphs through the feature filter based on the gated switch attention network to obtain a sequence of training purified sample centrifugal state time series correlation feature vectors; Passing the sequence of centrifugal state time series associated feature vectors of the training purified sample through the centrifugal state reasoner based on prior information to obtain a training centrifugal state time series reasoning semantic feature vector; Correcting the training centrifugal state temporal reasoning semantic feature vector to obtain a corrected training centrifugal state temporal reasoning semantic feature vector; Passing the corrected training centrifugal state temporal reasoning semantic feature vector through the classifier-based centrifugal controller to obtain a classification loss function value; The sample centrifugal state temporal correlation feature extractor based on the three-dimensional convolutional neural network model, the feature filter based on the gated switch attention network, the centrifugal state reasoner based on prior information and the centrifugal controller based on the classifier are trained based on the classification loss function and through back propagation of gradient descent.

6. A proteomics sample pretreatment device, characterized in that: The proteomics sample pre-treatment device is used to perform the proteomics sample pre-treatment method according to any one of claims 1 to 5.